A novel population robustness-based switching response framework for solving dynamic multi-objective problems

In this paper, a novel population robustness-based switching response framework (PR-SRF) is proposed to develop effective dynamic multi-objective optimization algorithm (DMOA), which integrates different response strategies to comprehensively cope with the dynamic behaviors. In particular, the popul...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Neurocomputing (Amsterdam) Ročník 583; s. 127601
Hlavní autori: Li, Han, Fang, Zheng, Hu, Liwei, Liu, Haonan, Wu, Peishu, Zeng, Nianyin
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Elsevier B.V 28.05.2024
Predmet:
Dynamic multi-objective optimization algorithms (DMOAs)
Evolutionary transfer optimization (ETO)
Population robustness
Switching response framework
Evolutionary transfer optimization (ETO)
Switching response framework
Population robustness
Dynamic multi-objective optimization algorithms (DMOAs)
ISSN:0925-2312, 1872-8286
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In this paper, a novel population robustness-based switching response framework (PR-SRF) is proposed to develop effective dynamic multi-objective optimization algorithm (DMOA), which integrates different response strategies to comprehensively cope with the dynamic behaviors. In particular, the population robustness is described by the quantification of how severely the environmental changes affect current population, which is timely graded as three levels of weak, strong, and normal to enable the adaptive switch of three different responses of diversity-enhancement, diversity-maintenance, and the knowledge-transfer, respectively. In this way, associations between the adopted responses and the changing environments are successfully established, thereby facilitating more intelligent decision when handling the dynamic behaviors. According to the benchmark evaluation results, the proposed PR-SRF-DMOA yields better comprehensive performance than several other DMOAs with popular response strategies, and it also outperforms another three DMOAs with hybrid responses, which demonstrates the great competitiveness of our algorithm. In addition, ablation study proves that the proposed PR-SRF can sufficiently exploit the merits of different responses, which effectively alleviates the negative knowledge transfer in extremely fluctuating environments, thereby providing valuable references for the development of evolutionary transfer optimization (ETO) algorithms.
ISSN:0925-2312
1872-8286
DOI:10.1016/j.neucom.2024.127601